
You use the hit list comparison tool to identify differences in the hit lists of two traces. This functionality is useful if you want to check your programs for nonlinear coding. You trace a program twice, the first time with a smaller amount of data, the second time with a larger amount of data. You use the same system and you do not change the program code between the measurement of the two trace files.
You also use the hit list comparison tool to see the effects on runtime if you change the code or the system. In this case you do not change the amount of data.
You have prepared two sets of test data for tracing.
We recommend using a reasonable amount for the first measurement, and a larger amount for the second measurement. An increase by a factor of 10 between the two measurements is most convenient.
Before you perform the actual tracing, you have first executed the program to fill the buffers.
We recommend the following measurement restrictions in a variant definition:
On the Statements tab, select DB-level Ops..
On the Duration/Type tab, select Aggregation by Call Position.
Data Loading and Checks
After you have started the Hit List Comparison tool, the system loads the trace files and checks whether the following conditions are true:
The trace name and type must be identical, otherwise an error message appears.
The aggregation type should be identical, otherwise a warning message appears, but the comparison is continued.
The selected trace events must be identical, otherwise an error message appears.
Whether for older traces the aggregation type is full. In this case, the comparison of internal tables does not work.
Whether the trace measurement is executed in parallel for only one of the two traces. In this case, a warning message appears.
Whether only one of the traces was restricted to particular units. In this case, the system tests the settings of the units and an error or warning message appears.
Comparison Process
After successfully loading the trace messages, a search is made in trace 2 for a line that corresponds to each line in trace 1. A match is only made when the following attributes are identical for both lines.
Event class
The top level of the event hierarchy.
Event Type
Event Name
The name is not checked only for internal tables whose names are not used.
Calling program
Program where the call comes from.
Offset
The line where the call is made. If the same event is called from two different positions in the code, the system regards it as two different events.
The hit list comparison tool automatically determines the order of the traces. The system by default assigns the trace with the larger net time to Trace 2.